Weightless Neural Networks are Artificial Neural Networks based on RAM memory broadly explored as solution for pattern recog- nition applications. Due to its memory approach, it can easily be im- plemented in hardware and software providing efficient learning mecha- nism. Unfortunately, the straightforward implementation requires a large amount of memory resources making its adoption impracticable on mem- ory constraint systems. In this paper, we propose a new model of Weight- less Neural Network which utilizes Bloom Filters to implement RAM nodes. By using Bloom Filters, the memory resources are widely re- duced allowing false positives entries. The experiment results show that our model using Bloom Filters achieves competitive accuracy, training time and testing time, consuming up to 6 order of magnitude less mem- ory resources in comparison with the standard Weightless Neural Network model.

--

• Computer and Information Sciences - Natural Sciences